摘要
复杂、并发故障诊断的难点在于这些故障的振动信号很复杂 ,特征很难获取。文中阐述了从实例数据库中挖掘故障特征的整体结构 ,定义了信号的绝对、相对和梯度特征及相应的绝对、相对和梯度模式实例 ,进而阐述了应用模糊聚类分析挖掘特征模式的方法。最后以往复式压缩机实例挖掘系统为例说明了该原理的应用。
The difficult problem of complex and syndrome fault diagnosis is that the diagnostic features might not be obtained because the signal of the fault is very puzzling. This paper proposes a diagram of the system of data mining diagnostic patterns, defines an absolute, relative and gradient diagnostic parameters and correspondingly patterns of absolute, relative and gradient fault cases. The method of data mining diagnostic patterns using fuzzy clustering is discussed. In the last, the diagnostic pattern data mining system from fault case database of reciprocating compressor is demonstrated.
出处
《振动工程学报》
EI
CSCD
北大核心
2002年第3期337-342,共6页
Journal of Vibration Engineering
基金
国家"九五"攀登计划资助项目 (编号 :PD95 2 190 8Z1)
河南省科技攻关项目资助 (编号 :2 0 0 0 112 0 32 3)
关键词
振动信号
故障诊断
数据挖掘
特征提取
知识获取
实例学习
fault diagnosis
data mining
characteristic obtaining
knowledge discovery
machine learning
